Why no love for Bancroft? He was an unbelievable defensive shortstop (at least according to Win Shares). He was an average hitter for a hitter, a very good hitter for a SS, with peak OPS+ 119-120 for 3 seasons. His career wasn't long, but it wasn't short. I think he's an easy choice for eventual induction, I guess maybe I'm missing something? What's not to like here? Is he just flying under the radar?

It seems to me that Dr. Chaleeko's post 74 on the ballot thread raises the question that I'd like to see supporters of Sewell and Bancroft answer.

What sets either of these guys apart from the collection of very good infielders who have been falling short of the ballot over the last ten elections or so? What makes Sewell or Bancroft better than Herman Long, Johnny Evers, Joe Tinker, Dobie Moore?

Everything Joe has just said about Bancroft, for example, could be said about Joe Tinker, couldn't it?

What makes Bancroft and Sewell as good or better than Larry Doyle or Cupid Childs?

I'm willing to be convinced that we all ought to be trying to sort out this bunch of infielders and putting the best two on the ballot in competition with the borderline pitchers and outfielders who are serious candidates for election now, but I need to see the argument made.

Evers, Tinker, and Long all fell out of sight. Childs and Doyle lack the support needed to make them serious candidates. Aside from a handful of strong supporters of C & D, there seems strong agreement that these players all fall short of what's needed for election to the HoM.

Bancroft and Sewell look to me like players of very much the same calibre as these five (and Dr. Chaleeko's numbers back that up). My system evaluates all of them, consistently, I believe, as not strong enough to make the ballot.

Possible conclusions:
1) My system is right -- none are ballotworthy.
2) My system is wrong -- all should be serious candidates.
3) My system is wrong -- Bancroft or Sewell or both are better than these other guys.

What's the evidence that 2 or 3 is the right conclusion rather than 1?

p.s. My system rates players within periods first, so my finding is not only that these players are similar when compared directly to each other, but that their rank against their contemporaries is similar also.

To coattail Kelly in SD's please for care and to work a little more through a point I wrote in the discussion thread....

Before we get excited about these new guys, someone show me why any of them would be a better selection than Herman Long.

Here's some interesting WARP information about them:
Long
WARP1: 109.4
WARP3: 67.5
diff: 38% discount

Bancroft
WARP1: 111.9
WARP3: 76.1
diff: 32% discount

Sewell
WARP1: 103.8
WARP3: 88.9
diff: 14% discount

[note: all WARP numbers as of latest iteration]

I see two stories in these numbers:
-One: Bancroft's NL is discounted at a rate about double that of Sewell's AL, despite the fact that they were contemporaneous for a large portion of their career.
-Two: Long's league context is discounted at a rate 18% lower than Bancroft's and at a rate more than 2 and half times that of Sewell's.

I may simply be using an incorrect and overly simplistic means to look at this question (please, someone say so if I am!). Still...

Is this a reasonable "discount structure" to describe the timeline differences between these three candidates? And if it's not, where is Long in this discussion?

Here's another take on Sewell/Bancroft:

Who was the better offensive player? Without a doubt, it was Sewell.

Who was the better defensive player? It was Bancroft, but Sewell was not that far behind. From my reading, Sewell had an excellent rep as a defensive player, but just misses the "Oh, my god!" greatness of a Bancroft, Smith or Robinson.

Despite the fact that Sewell played less seasons, he actually had more PA, which leads me to:

Who had the better extended peak? Sewell.

How about competition? I think Sewell played under slightly better conditions. I'm not convinced that it is any where near as severe as what BP suggests, however.

I don't think there is a huge chasm between the two players, but when you have players who are not as distinguished as a Cobb, Wagner or Lloyd, on the ballot, it's reasonable to have one high up on your ballot while the other is left off.

FWIW, I might move Sewell into the second or third spot after further examination.

I understand that, in a wide open election as 1939 is shaping up to be, that getting it right is more important than it was in '34. However, who ever is elected, I'm confident I won't be holding my nose while I construct their plaques (even if he didn't make my ballot). IOW, chill (except when Pearce was eligible, of course)! :-)

Warning: Mild-mannered anti-shortstop rant ahead:
------------------

Okay, I'm not one to fetishize positional balance, but at some point I think there has to be a step back and second look.

By my rough count, assigning each HoMer to a single position, there are currently 10 shortstops and 10 left fielders in the HoM. Now, that is not to say that we shouldn't vote for Arky Vaughan or anything because there are too many shortstops, but maybe we should stand back, look at the ten shortstops in the Hall, the three above, and ask where they fall in the Top 13.

Here it is, roughly, by WARP-1, (this is not my personal ranking, by my personal rankings are interspersed through my insertions of the relative rank of players who are not fully addressed by WARP -- early players and Negro Leaguers). Eligible non-HoMers are in bold:

1. Honus Wagner: 232.6
2. Pop Lloyd
3. Bill Dahlen: 171.1
4. George Davis: 159.4
5. Home Run Johnson
6. George Wright
7. Bobby Wallace: 154.9
8. Jack Glasscock: 138.2
9. Dickey Pearce
10. Rabbit Maranville: 132.8
11. Monte Ward: 120
12. Joe Tinker: 113.2
13. Dave Bancroft: 111.9
14. Herman Long: 109.4
15. Joe Sewell: 103.8
16. Tommy Corcoran: 101.8
17. Donie Bush: 102.5
18. Art Fletcher: 91.7
19. Hughie Jennings: 89.6
20. Ed McKean: 83.3

The problem, here, is obvious. Maranville's all career and no peak. Jennings is all peak and no career. And between them are a bunch of guys who are practically indistinguishable. Are any of the three names in the thread better than Joe Tinker? He racked up more WARP than anyone but the Rabbit, in fewer games than Bancroft or Sewell.

Win Shares has the order of these guys reversed: Sewell (277), Bancroft (269), Long (265), and Tinker (258), but we know that WS limits fielding shares in ways that are hard to predict. On the other hand, divide by three to convert to WARP-1 equivalent and they are all 10-15 wins worse.

In my view, all four of these plus, plus "no peak" Maranville and "all peak" Jennings are approximately identical, which is about what you expect when you start slipping down the bell curve from the Top 10 into the 11th through 16th best ever at a position.

And it's also the reason that I'm not voting for ANY of them.

Let me try another list at another position:

Name: WARP-1 (WS)

1. Cap Anson: 164.5 (381)
2. Roger Connor: 152.4 (363)
3. Dan Brouthers: 130.5 (355)
4. Joe Start
5. Jake Beckley: 113.5 (318)
6. Ed Konetchy: 101.9 (287)
7. Fred Tenney: 92.3 (249)
8. George Sisler: 88.6 (292)
9. Jake Daubert: 80.8 (263)
10. Stuffy McInnis: 77.4 (227)
11. Harry Davis: 73.7 (238)
12. Frank Chance: 73.3 (237)
13. Wally Pipp: 67.2 (203)
14. Jack Fournier: 67.1 (231)
15. George Kelly: 67.1 (193)
16. Lu Blue: 65.5 (198)
17. Hal Chase: 65 (231)
18. Henry Larkin: 63.4 (177)
19. George Burns: 63.4 (200)

Personally, I like Beckley. Others, I see, prefer Sisler (the Jennings of the group). Ed Konetchy might be the forgotten man, with the intermediate peak and career values that excite neither fan, but might be more valuable than either. Frank Chance was also popular once.

Along with considering why Tinker isn't better than the 3 shortstops mentioned above, I nominate Ed Konetchy, who has more Win Shares than any shorstop except Maranville, a comparable WARP-1, and would definitely a Top 10 first baseman if inducted. (The shortstops would all be #10, at best, if inducted).

My personal ranking of only first basemen and shorstops:

1. Beckley
2. Sisler
3. Jennings
4. Konetchy
5T. Sewell
5T. Tinker
7. Bancroft
8. Chance
9. Maranville
10. Long

Presently, only #1 (Beckley) makes my ballot. Either Jennings or Sisler (they are close) may some day, if they aren't inducted first. The rest -- well -- the implication of this thread implies that if we could just figure who was the best of the three, we could induct him. In my mind, none of them are really close.

Phillybooster:

Great post. One question:

Where would you put Ben Taylor and Dobie Moore?

Good question. One of the flaws in my analysis process is that when looking at Negro League players, they either get slotted in the Top 20 (currently Mendez, Monroe, Redding) just get dumped out, so aren't there when I go back to re-analyze unless I go back and look for them specifically again.

Taylor, I would slot between Konetchy and Sewell (call him 4A), and would be in the low 20s. Moore I'm not so hep on, what with the short career and shorter peak. He'd be out of the Top 10. Looking back, again, Sol White would also be 8A on my list.

Moore has an est career WS total of 245-255, dividing by three that's 82-85 WARP.

That puts him at the bottom in the McKean/ Jennings range careerwise.

Of course, if you look peak/primewise, he's at or near the top of the list among the bold-faced eligibles (save for Jennings, of course).

PhillyBooster:

Regarding shortstops, it's not as severe as you have presented it since Pearce represents a decade (the 1860s) that hardly has anyone who represents it (Start would be the other one), while Ward is a hybrid.

With that said, I agree we should always step back to make sure we're not overrepresenting a position.

Question: Did Sewell play when shortstops were poorer hitters? Answer: yes, by about 3 runs per year, compared to the other two.

Players careers, chopping off very poor seasons at the beginning or end of their careers:

shortstop years .PA. .OWP SSavgOWP RCAP
Bancroft.. 15-27 7231 .524 .426 ...165
Sewell.... 20-33 8330 .549 .396 ...346
Maranville 13-30 8653 .429 .427 ... -6

But still, Sewell was a better hitter than Bancroft, and more durable. Even if you take 3 runs/year and add them to Bancroft's RCAP, Sewell beats him by about 155 runs. I don't think Bancroft was near 12 runs/year better with the glove.

Maranville was a superior glove, but his stick was waaayy worse than Sewell's.

RCAP (offense only) on the guys from the list above, and again, I toss out rotten years at the end so guys aren't dragged down by those:

Rabbit Maranville -6
Joe Tinker 23
Dave Bancroft 165
Herman Long 105
Joe Sewell 346
Tommy Corcoran -300; below avg Every year!
Donie Bush 106
Art Fletcher 93
Hughie Jennings 276
Ed McKean 200

Sewell has by far the best career stick for a shortstop. AND he was one of the better fielders as well.

Regarding shortstops, it's not as severe as you have presented it since Pearce represents a decade (the 1860s) that hardly has anyone who represents it (Start would be the other one), while Ward is a hybrid.

But doesn't that make it MORE severe?

If there are 10 SS and 4 1B, then SS is 250% better represented. Subtract Pearce and Start and you're down to 9 and 3, or 300% better represented (or 283%, if you cut Ward in half).

With that said, I agree we should always step back to make sure we're not overrepresenting a position.

It's the counterargument to the oft-heard "We are comparing to other eligibles, not people already inducted." That is true, but how do we know how to compare them if we do not know what scale to use?

It is 1939, and the last of our most contemporary HoM first basemen retired 42 years ago. Positions go through ebbs and flows, but unless we are going to conclude that first base was the early 20th century equivalent of DH (which it certainly was not), then I do not see how your favorite shortstop left on the ballot (whoever he is) goes above your favorite first baseman.

then I do not see how your favorite shortstop left on the ballot (whoever he is) goes above your favorite first baseman.

That would be if you honestly don't think any of the first basemen were that good. I'm all for position balance and everything, but I personally am not voting for (in my view) mediocre players in order to make our position list look nice and balanced.

It is certainly possible that the top 2 or 3 eligible shortstops from the 20th century would be better than the best first baseman, but the top 6, 7, or 8?

It is then that I find the balance shifting more away from "my analysis shows yet another Shortstop X is better than First Baseman Y" and toward "my method of analysis is flawed".

If you look at it decade by decade, in 1880, 1890, and 1930 there is more HoM talent playing 1B than SS, while in 1870, 1900, 1910, and 1920 there is more HoM talent playing SS than 1B. What goes around comes around.

That having been said, if you go decade-by-decade there is a case to be made that Bobby Wallace was overkill and/or that one or another of Beckley, Chance, or Sisler should eventually be inducted. For the moment, I'll be agnostic on the Wallace question and chalk up the failure of 1B to achieve adequate representation during its down decades to the inability of Frank Chance and George Sisler to stay healthy.

The health luck turned pretty resoundingly against the shortstops after 1920, of course, with Ray Chapman and Dobie Moore.

Yes there are lots of SS in the HoM, but many of the greatest athletes have played SS especially in pioneer type environments (19th century, Negro leagues). I have no problem electing more if they're the best candidates.

But starting with the 20th century SSs who ought to be fairly easy to sort out. I use a series of head-to-heads when we get into the non-NBs.

1. Bancroft and Sewell. Beauty had a slightly better 3 year peak by WS, WARP and TPR; Sewell a better 5 by WS and WARP and it's a dead heat by TPR.

For career adjWS, I have Sewell 277-276, but WARP and TPR give Bancroft a significant advantage.

As for prime--Bancroft's was 10-12 years, depending on the method; Sewell's 9-13, and at a very slightly higher rate.

These guys are really really close, but if you believe Sewell faced tougher competition, that is a tie-breaker. Sewell by a nose.

2. Bancroft and Tinker. Bancroft had a better peak on every one of 8 measures I use. On many, it's close, but 8-for-8. Career it is also Bancroft on every measure. For prime, Tinker's is 13 years to Bancroft's 10-12 but Bancroft has at best equal or a higher rate. Bancroft by a head.

3. Tinker and Maranville. TPR prefers Maranville's peak, otherwise it's Tinker on both WS and WARP. For career it is pretty much all Rabbit. For prime, Tinker's 13 year prime is at a higher level than Rabbit's 9-11 year prime, except on TPR. Tinker by a length.

For the 19th century it is of course Jennings and Long. Among the Negro Leaguers it is Dobie Moore and Pelayo Chacon. And remember, Chris, Moore's career was cut short, but not that short. 14 years of elite ball, which is more than Bingo DeMoss or Ollie Marcelle.

Now what?

1. As a peak voter, the first head-to-head is Jennings and Moore. And remember Moore played 7 years in the NeL and also 7 years for a top barnstorming Army team. His was not a significantly short career for a pioneer type of environment. However, Jennings faced tougher competition. Jennings by a length.

2. Sewell (best of the 20th century) versus Moore (using Jennings as his surrogate, if you follow me). Jennings' peak is anywhere from 20 to 60 percent better than Sewell's by various measures. Jennings' prime rate similarly is 25 to 75 percent better, but his prime of 5-7 years versus Sewell's 9-13 is only 55 to 85 percent as long, so that Sewell has the edge on most career numbers but only 106-105 on adjWARP1. Jennings by two lengths, so that by comparison Moore also rates ahead of Sewell.

3. Sewell versus Long. This is tough. Long has a better peak on adjWS and WARP, especially since his best years came before 1893 when I give 50 percent of pitching WS to the defense, which probably overstates it. Still, Long has the better peak. Long also has a slightly longer prime on balance, though I see Sewell at a higher rate. For career Long leads on WS and WARP, but Sewell by a laughable 36.4-6.0 on TPR. Still, overall I see Sewell by a nose.

4. Bancroft and Long. A verbatim repeat of the Sewell-Long match-up. Bancroft in a photo finish because of a little better consistency. Long had a higher peak and a longer decline.

5. Long versus Tinker. All of the same dynamics apply here as to Sewell and Long. In this case I see it the other way around, Long by a nose because that higher peak is, I think, more real in this case. i.e. Long hit pretty well at his peak while Tinker's value, like Bancroft's, was much more on the defensive side. Long by a nose.

6. Tinker or Chacon, and Maranville or Chacon. Guesswork, but I like both of the major leaguers over Chacon. Chacon projects to about .250 with a Gold Glove and played 22 years, however, so I think he belongs on the list of good-field, so-so hit SSs with Joe and Rabbit.

In summary:

1. Jennings--this is easy
2. Moore--vastly better peak than anybody below and a 14 year elite career
3. Sewell--very close, but faced better competition than the rest
4. Bancroft--all-time defense, very consistent
5. Long--packaged offense and defense together at his peak more than the rest
6. Tinker--best of the three poem guys
7. Maranville--a lot of those 300WS was just hangin' around time
8. Chacon--not to be forgotten

And then just for fun:
9. Fletcher
10. Bush

But in a year or two, Dick Lundy will blow all of these guys away (OK, except Jennings). But if anything this exercise has reminded me just how fabulously good Dobie Moore was, and not just for 7 years. But not as good with the glove as Hughie.

Then if you slot these SSs in with IFers generally, without the explanations, here's where I'm at.

1. Jennings
2. Monroe--nobody on this list beats the Pearl on both peak and career. The high peaks had shorter careers and the longer careers had low/no peaks.
3. Moore
4. Sewell
5. Williamson--comps Jimmy Collins
6. Childs
7. Bancroft--very close vs. Doyle
8. Doyle
9. Dunlap--might have been better than Childs
10. Long
11. Tinker
12. Evers
13. Maranville
14. Leach--surprising to me that he comes out this low... again; similar peak but lower prime rates than Tinker; similar prime but lower peak than Evers; then the positional shift just makes him hard to digest
15. Sol White--there might be other MLers who rate ahead of these four, but 12 MLers on this list is plenty, and I wanted to be sure I gave the NeLers fair consideration.
16. Chacon
17. Marcelle
18. DeMoss

Finally I need to figure out how to slot in the other Gold Glove candidate, Max Carey, on this list. And the pitchers.

At least the "hitters" are easier.

But doesn't that make it MORE severe?

If there are 10 SS and 4 1B, then SS is 250% better represented. Subtract Pearce and Start and you're down to 9 and 3, or 300% better represented (or 283%, if you cut Ward in half).

What I meant was, we weren't really going after players from the 1860s. There might be other quality players from the other positions that were worthy from that time (which would make the numerical difference between shortstop and the other positions smaller), but we're just not confident of how great they were.

2. Moore--vastly better peak than anybody below and a 14 year elite career

14 years? He only played for seven seasons. I agree he has the best peak, without a doubt (including Jennings).

But in a year or two, Dick Lundy will blow all of these guys away (OK, except Jennings).

I'm looking forward to the discussion.

Sunnyday,

Wow, I hadn't realized that Moore was born as early as he was, 1893. That puts the end of his career at roughly age 33 (says Mr. Math Wizard...) and his rookie year at 26or 27. I had thought for some reason that he debuted at around age 20 in the NNL. OK, I'm a dunderhead who didn't do his homework well enough.

I've been giving Moore three years' worth of MLE credit, but I think I may now consider giving him more, probably a fourth year, to place his entrance into the league around age 22-23. Given that among his rough contemporaries Sewell debuted at 21, Frisch at 20, Gehringer at 21, Vaughan at 20, Cronin at 19, and Beauty at 24 this seems emminently appropriate for a player who could clearly mash.

2. Moore--vastly better peak than anybody below and a 14 year elite career

14 years? He only played for seven seasons.

Forget about what I said. I forgot about the infantry years.

I need to take another look at him. I think I may have been shortchanging him. I know he's another shortstop, but... :-)

John, with all due respect, you haven't processed the Dobie Moore discussion. He played 7 years, age 19/20 through 26/27 on an elite black army barnstorming baseball team--the same that produced Heavy Johnson--that routinely pounded the tar out of everybody with the courage to take the field against them. Granted, these weren't white MLers he was beating, but it seems likely from the Dobie Moore thread that they were as good or better than many of the black teams Rube Foster pitched against.

After that, he played 7 years in the NeL, age 26/27 through 33.

Now, how many MLE seasons anybody wants to credit for his 7 years in the NeL is up to them, but it was seven PRIME years, his "rookie" year(s) having been taken up elsewhere. And how many MLE seasons anybody wants to give for his army play is up to them, too, and he was probably pretty darn rough for the first few years, though we don't know that. I remember Chris Cobb saying he would credit Dobie with 5 (???) years out of the 7 as being at least NeL equivalent.

But clearly he played 7 PRIME years in the NeL. If any NeL play at all is ever MLE, then all of Dobie's was. As for the army time, again, anybody's guess. But keep in mind, when he went into the army--either TO (for the purpose of) play ball, or he was immediately pressed into service on the primary black army baseball team--he obviously could play some, and it was not because he had been rejected by ML managers or even NeL managers, it was in part because nobody had seen him play. So we do not know that he was not playing at a level worth crediting here right from the beginning, unlike say Rixey or McGinnity. What we know is he didn't have the chance. But he ended up with the very best baseball team he could possibly have played on.

So anyway, I see him as having a longer peak than Hughie Jennings, though at a lower level even before the NeL discount (don't think his defense was as great); a longer prime than Jennings, though at a lower rate; but at least as much career value, and surely more until and unless you apply a NeL or other discount, which of course you probably should.

But then of course you have to be a peak voter to like him at all. But to say he played 7 years, period, end of discussion, misses all the subtlety of his case and his times.

PS. On his thread, there is a comparison of Moore's WS peak projection to those of Cronin and Banks, and nobody seemed to disagree with it. Of course his career totals are closer to Maury Wills.

Other than Jennings, though, you put peak and career together and his career shape looks something (not a lot, but something) like Lou Boudreau. Another HoMer suggested it looked like a mirror image of George Sisler--i.e Moore's peak coming later, versus Sisler's earlier in his career. (Except that Moore was a SS.)

There is a lot of guesswork as to what his comps really are, but on the Moore thread I was just reminded that among the teams his army buddies beat were a number of PCL teams.

There have been a number of players in HoM history who had a lot of support when first eligible and then slipped badly. Somebody recently asked what happened to Spot Poles. Unless you are a died in the wool career voter and don't care much about peak, Moore clearly rates ahead of Poles.

Maybe Ernie Banks really is a good comp with Moore's 7 NeL seasons comping for Banks' SS career, and Moore's army days comping for Banks career at 1B.

John, with all due respect, you haven't processed the Dobie Moore discussion.

Marc, with all due respect, you didn't read post #21. :-D

A quick refresher on Dobie Moore, since he is (deservedly), getting some renewed attention. This is quoted from Gadfly’s research, posted on the Dobie Moore thread:

The best available evidence (1920 and 1930 census research) has Walter "Dobie" Moore born in 1895. [1893 is the date given in Riley.]

Dobie Moore began his baseball career playing with local teams in his Atlanta home town. In 1916, Moore was recruited by the United States Army's 25th Infantry (one of four Black divisions in the Army) to play for their baseball team. This team was commonly known as the 25th Infantry Wreckers.

On the basis of this information, I credited Moore with three MLE seasons in the army (1917-1919), bringing his total MLE career to 9.5 seasons (he was shot midway through the 1926 season).

Is it possible that 1Bmen were generally less valuable (and meritorious?) during the "lean years for the position" as seen here, as many as 42 years. Yes, it is. Perhaps Baseball misunderstood the position and relied too heavily on the Hal Chases and Jake Dauberts, including short careers of relative unknowns with prospects like theirs.

John Murphy #9
Regarding shortstops, it's not as severe as you have presented it since Pearce represents a decade (the 1860s) that hardly has anyone who represents it (Start would be the other one), while Ward is a hybrid.

It may be worth asking the same question about 1Bmen: McVey, Stovey (. . . years later, Banks Killebrew and Stargell). The net result, I'm sure, will be to subtract fractional SS and add fractional 1Bmen.

Dr. Chaleeko #3 (recently earned a PhD?)
Before we get excited about these new guys, someone show me why any of them would be a better selection than Herman Long.

Here's some interesting WARP information about them:
Long
WARP1: 109.4
WARP3: 67.5
diff: 38% discount
. . .
Is this a reasonable "discount structure" to describe the timeline differences between these three candidates? And if it's not, where is Long in this discussion?

No, the analysis is incorrect. Long played shorter seasons (132 to 154 games scheduled) and that is incorporated in WARP3. Check the definition of WARP2.

not to say this undermines Doctor's point

I don't think you can directly divide WS by 3 to get WARP. WARP has too low of a replacement level for defense, so all of the numbers are too high, especially at the key defensive positions.

Thanks for the discussion guys. I had Beckley #1 in 1938, Bancroft 10 or 11 so I guess I'm on the same page as Philly Booster. I think 1B defense was much more important in the deadball era, so 1B didn't have to hit as well as they did after it to be productive.

I have to look at Sewell still, but I imagine I'll have him ahead of Bancroft. I had Dobie Moore ranked high earlier (I'm one of the only people that has actually voted for him) and I'm going to take another look at him as well.

On this ballot, I see myself voting Rixey/Faber/Beckley at the top and then a few of the older guys like Pike and Jones, followed by Bancroft, Sewell, etc. Should be an interesting ballot.

WARP has too low of a replacement level for defense, so all of the numbers are too high, especially at the key defensive positions.

And Win Shares has too low of a replacement level for offense, so all of the numbers are too high, especially at the key offensive positions. And Win Shares does not adapt at all well to changing defensive conditions.

I think 1B defense was much more important in the deadball era

Somewhat more, yes. Much more, no. First base men did not hit much less than outfielders during the "drought". They still hit considerably more than did any other infield position. Any extra credit should be given more to defensive 1b-men than to hitting 1b-men.

Average OPS by position and by decade:
Decad 1B LF RF CF 3B 2B Ca SS Pit
1870's +1 +4 -1 +4 +2 +2 +0 +1 -13
1880's 13 +6 +1 +5 +1 -1 -7 -2 -17
1890's +6 +9 +7 +7 +0 -2 -6 -2 -22
1900's +6 10 +9 +8 +0 +2 -9 -1 -29
1910's +6 +7 +9 10 +1 +1 -7 -4 -31
1920's +9 10 10 +8 -3 +1 -4 -7 -32
1930's 13 +8 10 +5 -1 -3 -3 -4 -36
1940's +8 11 +9 +7 +2 -3 -4 -4 -37
1950's +9 10 +7 +7 +4 -3 -1 -5 -40
1960's 11 +9 11 +7 +4 -5 -3 -6 -46
1970's 10 +8 +8 +5 +3 -5 -2 -11-45
1980's +8 +6 +6 +2 +3 -4 -4 -8 -48
1990's +9 +4 +6 +1 +1 -3 -4 -7 -50

FYI: In the early days, RF was the most common position for a substitute pitcher. Offhand, I don't know that any sub P commonly played C, 3B, SS or LF.

Beginning in 1892, given free substitution, there was no reason to play a substitute pitcher, as far as I know.

jimd,
What is the population base for your average OPS by position?

Jim, I agree with what you say about Win Shares offensive replacement level being too low. But that still doesn't mean you can just divide by 3. I would think you could divide by 3 if you adjust WARP for defensive replacement level being too low, and if you subtract 6-7 WS for a player playing a full 162-game season. Neither of them adapt to pre-deadball defensive conditions too well if you ask me, though WARP is much better. WS does fine with the 2B/3B shift, but neither give enough credit to pre-deadball 1B defense. Both probably overrate 19th Century right fielders.

"Somewhat more, yes. Much more, no. First base men did not hit much less than outfielders during the "drought". They still hit considerably more than did any other infield position."

That's semantics. I think it's much more, because now 1B do not hit less than outfielders now, so it moves them from the 8th most important defensive position to the 5th most, which I think merits a 'much'. Semantics, we probably agree on the degree to which 1B defense needs to be moved up, I just think it's 'much' and you think it's 'somewhat' - but it's probably roughly the same when we put a number on it.

I think that pre-deadball the distance for 1B above average in your table is roughly 2/3 to 1/2 of what it is post-deadball, which I consider significant. 1B pre-deadball is between 3B/CF post-deadball. That's a major difference from where it is today.

"Any extra credit should be given more to defensive 1b-men than to hitting 1b-men."

I disagree, I think most of a player's defensive value comes from just being 'there' (especially at catcher) and allowing a better bat in the lineup somewhere else though there are exceptions for extremely good/bad defensive play. While I'd give the better defensive first-basemen a slightly higher boost, I give most of my 'boost credit' equally among all firstbasemen.

I also wouldn't use 'average' for the table above. I'd use an average of the bottom 3 regulars (meaning 14-16 overall for the pre-expansion period) at each position (replacement level). If not, a star-glut or drought at a position weighs too heavily in the totals. Like 1B in the 1930s or SS in the 1970s.

I'd imagine it'd be a ton more work though - it's a great table, but that's the tweak I'd make if I had the time. But a star glut doesn't signficantly change the replacement level, because the difference between the 16th, 14th and 12th best player (the one that is at replacement level depending on the number of stars) generally all that significant.

What is the population base for your average OPS by position?

The Lahman database. I divvied every players batting line proportional to his fielding games.

Yes, 1B was 5th on the defensive spectrum, but not by much. OTOH, there is a 'chasm' between the 'bats' (OF-1B) and 'gloves' (2B-3B-SS-CA). The average OPS of a 1B was consistent over the 30 years (1890-1920); their defensive credit is probably that appropriate to a modern RF, no more, though that's more than Win Shares grants them.

I think most of a player's defensive value comes from just being 'there' (especially at catcher) and allowing a better bat in the lineup somewhere else

Sorry, but I'm trying to figure out what this means.

I'd imagine it'd be a ton more work though

Unfortunately, it is. The database does averages of large amounts to data quite nicely.

I also wouldn't use 'average' for the table above. I'd use an average of the bottom 3 regulars (meaning 14-16 overall for the pre-expansion period) at each position (replacement level). If not, a star-glut or drought at a position weighs too heavily in the totals. Like 1B in the 1930s or SS in the 1970s.

I'd imagine it'd be a ton more work though

Yes, it is, and not principally because of limitations in the database. It is a ton of work, perhaps without hope of satisfying outcome, merely to identify regular players, one for each fielding position, each team.

The database can be used to estimate a total batting statistical line for each team-season-fieldpos; and, in turn, the 14th, 15th or 16th best team batting line at each fieldpos. (A)

It is also possible to select the worst batting line for anyone who played a threshold such as 75% of scheduled games played. That may be interpreted as batting record of the worst "regular" within a group of "regulars" that varies from fieldpos to fieldpos and from year to year. (B)

(B) generally overestimates replacement-level batting --almost everyone will agree, I think.

What about (A)?

"I think most of a player's defensive value comes from just being 'there' (especially at catcher) and allowing a better bat in the lineup somewhere else"

"Sorry, but I'm trying to figure out what this means."

What it means is that quality of defense is much less important than position played. Assuming equal offense, a great centerfielder generally isn't as valuable as a poor catcher, for example.

The poor fielding catcher allows a replacement level offensive CF (who is average defensively, since defensive replacement level is average) into the offense. That player is typically a much better hitter than his replacement level offense, average defense catching counterpart. So much better as a hitter that it negates the defensive advantage the good CF gives you, as opposed to the average catcher.

**********

Pre-1920 1B may have been 5th, by 'not much', and in the group closer to the OFs, but it's still the same amount over '0' as a player inbetween 3B and CF would be today. WARP and WS do account for that, and because of this, they underrate 1B from before the deadball era, by quite a large margin. By about as much as if you would if you consider a 3B/CF today a 1B.

**********

I actually think both A and B are reasonable. I think A would work out very well. Maybe using 50% for B would work too - but you'd have to be careful if games were your threshold, or you could end up with a bunch of defensive replacements as your data set. Maybe someone with 50% of an average season's PA would be better?

What it means is that quality of defense is much less important than position played. ... defensive replacement level is average

The model you are employing is that:
1) a separate offensive replacement level is determined independently for each position
2) defensive value is calculated relative to average at the position

This is not the model used in either WARP or Win Shares. Their model calculates one offensive replacement level for all positions. Win Shares and WARP have defensive systems that are very complex, taking into account overall team defense, etc.

I am not familiar with working with your model, and will have to think over how it compares with their models. At first glance, it seems like it follows the same theoretical lines as Palmer's Linear Weights. If it doesn't incorporate team defense somehow, I'd be surprised if it works any better.

By about as much as if you would if you consider a 3B/CF today a 1B.

Not by a long shot. Define modern as the last 30 years (1970-2000). The modern 1b-man is at +9; dead ball 1b-men are at +6, a 3 point difference. Modern CF'ers are at +3, modern 3b-men are at +2, twice or more the difference (6 and 7, respectively). Dead-ball 1b-men hit about half-way between modern 1b-men and modern 3b-men or cf-ers.

IOW deadball 1b-men hit like modern RF-ers. They were hitters first; they needed better hands than modern 1b-men because the old gloves didn't even out that skill, but there's little evidence that they should be considered as a real defensive position. It's just a little more difficult than OF, with apparently a higher injury rate due to the rough play around the bases.

JimD,

Are you sure that WARP doesn't have a by-position replacement level? I may misunderstand what they are up to, but I think they adjust a player's EQA for position before they put it through the WARP formula.

To use a current-year example that's on the site now (so Fall 04 version), compare teammates Darin Erstad (1B) and Adam Kennedy (2B):

EQR
Erstad 68
Kennedy 63

OUTS
Erstad 357
Kennedy 354

EQR/OUTS
Erstad .19
Kennedy .17

EQA
Erstad .263
Kennedy .264


If WARP didn't assess each position's offensive replacement level differently, then I think Kennedy's EQA would come out lower than Erstad's, not higher, since they play in the same ballpark and EQR are park-adjusted.

I did it again. Grrrrrrr! Erstad has 68 UEQR, I read the wrong column. This pretty much invalidates everything I just wrote so just disregard my previous post.

Although its a semantics issue primarily, I think the discussion of offensive and defensive replacement levels leads to some lousy thinking. Replacement level is not an offensive or defensive issue (with a few exceptions I will discuss below) but an entirety issue. A shortstop with below "offensive replacement" level that fields like Ozzie wont be replaced. Why? Because his total value is above replacement level. Replacement applies to the entirety of the player. A Shortstop has a single replacement level, not two different levels. When a shortstop is replaced, you are replacing both his offense and his defense.

Exceptions:
DHs (trivial exception)
Pinch hitting
Late inning defensive replacements

Splitting the replacement level in to two parts is probably useful for these last 2 situations. Bad hitters/good gloves get less at bats because they are pinch hit for. Vice versa for good stick/bad glove types.

As an example, I wish warp calculated BRAA, FRAA and RAPPT (Replacement Adjustment by Position and Playing Time). They split RAPPT into 2 parts (3 with pitching). I think that is mostly silly.

Replacement level is not an offensive or defensive issue (with a few exceptions I will discuss below) but an entirety issue.

Well said, robc. I've tried to make this point before in discussions of this topic with Joe. It's one of my complaints with WARP.

Yes, very important point.

BP's WARP methodology, unless they've changed it recently, will show a SS with exactly replacement level offense (0 WARP) and slightly above replacement level defense (2 warp?) as being 2 WARP overall, when in reality he would likely be a AAA SS, with a "real" overall replacement level of negative something...

"BP's WARP methodology, unless they've changed it recently, will show a SS with exactly replacement level offense (0 WARP) and slightly above replacement level defense (2 warp?) as being 2 WARP overall, when in reality he would likely be a AAA SS, with a "real" overall replacement level of negative something... "

I agree with this. Been harping on it for at least a year. That's what I mean when I say that WARP uses too low of a replacement level for defense. Defensive replacement level is average.

I still think they ARE two separate issues. If you look at fringe players, they are generally average defensive players that have poor offense relative to the typical player at their position. Most managers historically have preferred the hitter to the fielder, assuming the overall packages are equal.

Again this may be semantics. Of course there is a single 'entirety' replacement level. But there isn't any way to measure it that way. You have to measure the offense and the defense seperately and combine them.

When you combine them you have two choices. You compare a player first to all hitters. Then you compare him to fielders at his position and build in a 'floor' or 'scale' for each position and give a fielder credit based on that scale. This is what Win Shares does. I believe this is what WARP does too (since all hitters are rated on the same scale, not position specific, at least for EQA).

I still don't see how what I've said about defensive value doesn't concur with this. I think the intrinsic ranges at each position should be fairly limited - which is what I say when I mean that being 'there' is more important than being 'good' - within reason. Someone else may think that the ranges should be wider. This is one way that I 'tweak' the main systems (WARP and WS).

Win Shares is absolutely constructed this way. There is no way that a Gold Glove Firstbaseman could ever be considered as valuable as an average shortstop (in the same number of innings) by Win Shares. The scale for 1B is generally 1-4 WS per season. For SS it is something like 4-10 WS (again there are rare exceptions for awful or incredible defensive play).

**********

Looking at your table rather quickly Jim - I saw the 5, 7, 7, 7, 5 for CF from the 1930s-70s and the 2, 4, 4, 4, 3 for 3B from the 1940s-80s and threw the +6 in the middle. So I'll tweak my comment to say that a 1B from pre-1920 is generally comparable to a 1930-79 centerfielder defensively. I think the premise still holds. He was somewhat less defensively valuable thank a 1940-89 third-baseman, but still more valuable than any corner outfielder. It also looks like all positions moved closer to the average in the 1990s, which is kind of strange. I wonder what caused that.

While Dr. Chaleeko's example isn't a good one, I still believe that WARP sets replacement level by position. BP's RARP is runs above replacement at POSTION, and VORP is Value over replacement at POSITION. If they didn't do this, then SS's, C's, etc. would be vastly underrated.

Think about it, you have a first baseman and a shortstop. the 1B is a better hitter being say, 50 runs above replacement and the SS is 30 runs above replacement. Both are 10 runs above replacement with the glove. Is the 1B really 20 runs better? Isn't he more like 7 or 8 runs better of something like that since a replacement level SS is still contributing more than a replacement level 1B?

I guess it is just that since they run RARP and VORP by position, I figured they run WARP that way as well. Win Shares is completely different though since it doesn't really have a replacement level. It recognizes that a SS generally contributes more defensively and gives him more credit.

Defensive replacement level is average.

That depends on the model employed. If you have ONE offensive replacment level for all positions, then defensive replacement level is definitely NOT average. If you customize the offensive replacment level by position, then defensive
replacement level MIGHT turn out to be average.

An average fielding SS at the league-wide offensive replacement level is a much more valuable player than an average fielding 1B at the league-wide offensive replacement level. An average fielding SS at the SS-only offensive replacement level might be about as valuable as an average fielding 1B at the 1B-only offensive replacement level.

I draw this distinction because both WS and WARP use a league-wide offensive replacement level. Therefore they cannot utilize the concept of "Defensive replacement level is average" in any practical way, because it is not relevant to their model. Both systems employ the defensive spectrum and give more fielding credit to below average SS's than to above average 1B's.

Of course there is a single 'entirety' replacement level. But there isn't any way to measure it that way.

Of course there is. Measure each component relative to average. Determine how much the "fringe" player is costing in runs relative to the average player in each component. As a manager you want to minimize the total cost, determining which marginal player costs the least amount of runs relative to average. The worst such player playing regularly is presumably the "replacement level player". Determining the effective component replacement levels from the "entirety" replacement level may be more of a challenge.

On extreme teams (extreme offensive or extreme defensive teams) the true cost may be different from the "linear" cost due to distortion caused by Pythagorean effects. A great offensive team with poor defense will find that marginal defensive runs are more valuable than marginal offensive runs. A great defensive team with poor offense will find that marginal offensive runs are more valuable than marginal defensive runs.

******

OK. We're arguing different definitions of modern. I consider yours to be potentially "misleading" (though not intentionally so) because over the last 30 years, CF has evolved from what was definitely a hitting position to much more of a defensive position, maybe even swapping with 3B. If you want to equate deadball 1B-men with 50's 3B-men and CF'ers (like Mantle, Mays, Mathews), I'll quibble that deadball 1b-men hit a little more, on average. However, 3B-men and CF'ers of the last 25 years are much more defensive (or sorely lacking in such outlier stars) and only hit slighlty above league average, and should not be confused with dead-ball 1b-men.

It also looks like all positions moved closer to the average in the 1990s, which is kind of strange.

This may just be expansion reducing the impact of the outlier stars, though maybe something else is also going on.

Joe,

When you have been saying the defensive replacement level is too low, what you really mean is the WARP overall replacement level is too low. It all comes down to what record you think a team of all replacement level players would achieve. Are the 62 Mets and 03 Tigers replacement level? Or is it the Cleveland Spiders of 99? (crap, forgot the year). The warp replacement comes in at Spider level not Tiger level. An adjustment at about 1.5*games/tmgames should handle that difference (for position players).

jimd #43
OK. We're arguing different definitions of modern. I consider yours to be potentially "misleading" (though not intentionally so) because over the last 30 years, CF has evolved from what was definitely a hitting position to much more of a defensive position, maybe even swapping with 3B.

> [JoeD] It also looks like all positions moved
> closer to the average in the 1990s, which is
> kind of strange.

This may just be expansion reducing the impact of the outlier stars, though maybe something else is also going on.

In jimd's narrow modern era, pitchers bat and support high + scores at several other positions only in the NL. The pitcher share of MLB plate appearances may be down near 40% of its magnitude pre-DH (perhaps 1946-1972, broadly "modern").

For some purposes, the batting data may be more instructive if pitchers are removed.

robc #44
Joe, When you have been saying the defensive replacement level is too low, what you really mean is the WARP overall replacement level is too low. It all comes down to what record you think a team of all replacement level players would achieve.

Not only that. I daresay, what some people mean and understand by replacement-level fielding is fielding skill that makes a replacement-level player when combined with batting/baserunning that is average at that fielding position (or overall, or overall less pitchers).

On changing the replacement level. We know that team Win Shares are constrained to add up to a preset total (3xWins), and it's been conjectured that WARP-1 does too (expected wins). Changing the replacement level can have surprising results.

Take a group of 4 full-time players, values 3, 7, 10, 20, constrained to add to 40. Now let's change the replacement level and remove 5 units from each one
(-2, 2, 5, 15), and also maintain the constraint by rescaling the intermediate values to sum to 40. The new values are -4, 4, 10, 30.

The scrubs lose value (the lower the original value, the more they lose); there is a pivot point at which the entire operation had no effect (in this case, average, don't know if this holds true in general); the stars gain value.

In jimd's narrow modern era

It's about the same width as that of the "dead-ball first-base-man". It also corresponds to an apparent change in how CF is staffed. They don't hit like they used to, so comparing dead-ball 1b-men to 90's CF's is not appropriate. Should a new generation of slugging CF'ers appear soon, then the CF "drought" might be an aberration, or evidence of increased defensive requirements (perhaps due to the larger multi-purpose stadia).

What makes Bancroft and Sewell as good or better than Larry Doyle or Cupid Childs?

I think this is asking the wrong question for the eras we're still looking at.

If you're going to compare Doyle or Childs to other infield positions, it should be to FIRST BASEMEN and not to SS or 3B men.

OTOH, I think it is acceptable to compare Bancroft and Sewell to Home Run Baker, Heinie Groh, etc. as those two positions had some overlap during these time periods.

And actually, on most teams, the best 'leftside' infielder almost always played SS with the 2nd best 'leftside' infielder at 3B, so we should expect to induct more SS and few 3B from this time period in baseball.

"That depends on the model employed. If you have ONE offensive replacment level for all positions, then defensive replacement level is definitely NOT average."

When I say average, I mean average for the position. No matter what model is in play, unless Tango is wrong (or I'm interpreting him wrongly) an average defensive player at a particular position is replacement level.

This is my model and the WS model:

Offensive value (all players on same scale)
+
Defensive constant based on position
+
Defensive value

I think the 'defensive value' portion should be considered zero if the player is MLB average for his position defensively.

I think the offensive value number should be well above zero if the player is MLB average.

"An average fielding SS at the league-wide offensive replacement level is a much more valuable player than an average fielding 1B at the league-wide offensive replacement level."

Of course - that's where the defensive constant based on position comes into play.

"I draw this distinction because both WS and WARP use a league-wide offensive replacement level. Therefore they cannot utilize the concept of "Defensive replacement level is average" in any practical way . . . "

I disagree, I just showed how. Win Shares does this, though it may or may not set average as replacement level for the defensive value portion.

"The warp replacement comes in at Spider level not Tiger level. An adjustment at about 1.5*games/tmgames should handle that difference (for position players)."

I agree, though I think it's more like 2.5 not a huge issue though.

I see WS replacement level as around 7 WS for a full 162 game season - maybe the two are closer to 3:1 than I initially realized.

"They don't hit like they used to, so comparing dead-ball 1b-men to 90's CF's is not appropriate. Should a new generation of slugging CF'ers appear soon, then the CF "drought" might be an aberration, or evidence of increased defensive requirements (perhaps due to the larger multi-purpose stadia)."

I definitely agree that larger stadiums put more of a premium on CF defense, as opposed to 3B.

Which is where using replacement, not average to set the table would probably have a lot of value, so stars rushes and gluts wouldn't throw the table off.

My hypothesis is that this is an aberration and that other than 2B and 3B shifting importance from 1920 through WWII there hasn't been much of a change in the defensive spectrum since the deadball era. I'd be really surprised to see significant changes post WWII, other than maybe the CF-3B changes that were possibly influenced by late 1960s stadium construction.

"And actually, on most teams, the best 'leftside' infielder almost always played SS with the 2nd best 'leftside' infielder at 3B, so we should expect to induct more SS and few 3B from this time period in baseball."

But the 3B generally hit better, so I'd think it would be a wash . . .

jimd
It's about the same width as that of the "dead-ball first-base-man". It also corresponds to an apparent change in how CF is staffed. They don't hit like they used to, so comparing dead-ball 1b-men to 90's CF's is not appropriate.

Calling jimd's modern era "narrow" and JoeD's "broad" isn't intended as a criticism of either. jimd's point about average batting/running by fielding position must be sound (mental arithmetic shows) but its magnitude must be exaggerated and its details distorted by the AL use of DH during his modern era.

Average OPS by position and by decade:
Decad 1B LF RF CF 3B 2B Ca SS Pit
1870's +1 +4 -1 +4 +2 +2 +0 +1 -13
. . .
1960's 11 +9 11 +7 +4 -5 -3 -6 -46
. . .
1990's +9 +4 +6 +1 +1 -3 -4 -7 -50

I presume that +1 means
: fieldpos-average OPS = 1.01 x league-average OPS

because OPS is easy to calculate using a database that includes no sabermetric stats such as OPS+. Right?

Since the 1960s, the most recent decade pre-DH, jimd's table exhibits roughly the same large decreases in relative batting/running for all three outfield positions (see below). Why? Partly because AL pitchers no longer float large plus scores for all by depositing their own -50's two or three times per game.

1960s- -2 -5 -5 -6 -3 +2 +1 -1 -4 arithmetic difference
-1990s 1B LF RF CF 3B 2B Ca SS Pit (DH, PH not included)

>>
roughly the same large decreases in relative batting/running for all three outfield positions (see below). Why? Partly because AL pitchers no longer float everyone else by depositing -50 two or three times per game.

1960s- -2 -5 -5 -6 -3 +2 +1 -1 -4 arithmetic difference
-1990s 1B LF RF CF 3B 2B Ca SS Pit (DH, PH not included)
<<

Measured with base 1950s, there does appear to be a decrease in LF and CF batting relative to RF. But it appears that something happened in RF, if anywhere; the 1950s and 1960s were extreme decades for RF batting. jimd's table shows that LF and CF have moved in tandem since the 1880s except for the 1910s (LF weak; CF strong).

Offensive value (all players on same scale)
+
Defensive constant based on position
+
Defensive value

As it often turns out, we're arguing definitions here.

Let OV + DC + DV be the above equation.

(OV + DC) + DV is what you get if you adjust offensive replacement for position, and calculate a distinct replacement value for each position. In this calculation, defensive replacement is average.

OV + (DC + DV) is what you get if you calculate offense and defense independently. WARP presents the (DC + DV) term together as DRAR with DV being DRAA. Win Shares does something similar with (DC + DV) being Fielding Win Shares. The methods to calculate DV are based relative to average; the calculation of DC is based on the interaction between overall constants (e.g. SS is 18%), and the performance of the team defense (Fielding/Pitching split).

I don't think that the conclusion "defensive replacement is average" is implied here, because of that positional constant DC. You do.

"An average fielding SS at the league-wide offensive replacement level is a much more valuable player than an average fielding 1B at the league-wide offensive replacement level."

Of course - that's where the defensive constant based on position comes into play.

And this is exactly why I don't believe that "defensive replacement is average". Because average fielding SS's and average fielding 1B's that hit the same amount have different values, because the DC's are different, because when DV is 0, the OV necessary to have positive value is different.

****

I see WS replacement level as around 7 WS for a full 162 game season - maybe the two are closer to 3:1 than I initially realized.

Win Shares allocates roughly 1/3 of its value to pitching, and 2/3rds to position players; each of the 8 positions gets roughly 1/12 of the value. That is the basis of the following table, correlating the team record with the value of the average full-time regular:

WS -- Team record
 5 --  20-140  1899 Cleveland Spiders 20-134
10 --  40-120  1962 NY Mets 40-120, 2003 Tigers 43-119
15 --  60-100  typical cellar dweller (.375)
20 --  80- 80  typical average team (.500)
25 -- 100- 60  typical championship team (.625)
30 -- 120- 40  1906 Cubs 116-36, 2001 Seattle 116-46
35 -- 140- 20  1875 Boston 71-8 (doubled, 142-16)

Everybody has their own opinion of where replacement level lies. I've seen arguments made for replacement level being as high as .400. They've not convinced me, but if true then Win Shares replacement level could be as high as 15. If you believe it's at .250, then it's about 10, .300 puts it at around 12. .175 puts it at around 7.

Paul wrote:

Is it possible that 1Bmen were generally less valuable (and meritorious?) during the "lean years for the position" as seen here, as many as 42 years. Yes, it is.

It is certainly possible for a position to go through a long drought. It is conceivable that there are simply fewer worthy catchers than there are at other positions.

I understand a voter chosing a 300 Win Share player who was the 20th best first baseman ever over a 240 Win Share player who was the 15th best catcher ever. I may not agree, but I at least understand.

What I don't get here, though, is that voters prefer a player like Sewell to a comparable player like Konetchy who is higher rated at his position all-time, and has identical stats.

Ed Konetchy: 101.9 WARP-1; 287 Win Shares
Joe Sewell: 103.8 WARP-1; 277 Win Shares

Both had careers of similar length, with comparable peaks. Both were the best in their leagues several times.

Don't we start out with the assumption that these players are fairly identical, and should be about equal on our ballots?

Now, what if we throw in that Konetchy is currently the 4th-7th best first baseman in history, and Sewell is the 10th-16th best shortstop in history?

What am I missing? How is Sewell even close, let alone miles ahead -- looking to finish in the Top 10 while Konetchy struggles to make a handful of ballots?

Ed K's tiny advantages in career WARP and WS are based on his extra 300+ PA.

Most would agree that the 1920s AL was much stronger than the 1910s NL, which puts some space between them, especially on a tight ballot.

Off the cuff, Sewell his almost as well as Konetchy (by EqA or OPS+ or OWP), so we might expect a SS to outrank a 1Bman if the offense is close.

To call Ed K the "4th-7th best 1Bman" is a bit of a push. We've elected four, and Sisler/Chance/Beckley are in the mix, so let's say "5th-8th best".

But having said all of that, I'll take another look this week. Bill James said Ed rated much higher than he subjectively first looked, and part of it was his great D and his deadball-era stats, so maybe we are still not focusing right.

Sewell's OWP was .549 for his career.
The average player playing the same positions would have an OWP of .396.

Konetchy's OWP was .603 for his career.
The average player playing the same positions would have an OWP of .527

So Sewell was 39% better than his position(s), and Konetchy was 14% better than his.

Konetchy had about a 1/2 season more PA's.

Then there's defense, where Sewell was a very good fielding SS vs. Konetchy being an excellent fielding 1B - Advantage Sewell there also.

I don't see how Konetchy could be considered even close to Sewell in terms of value.

FWIW, pennants added using Win Shares has the two as dead even (.733-.732, Sewell ahead). Sewell had a couple of bigger seasons, but Konetchy had an extra season. Kontechy has more WS (303-293) but most of that gap is shaved off when compared to replacement level (214-212).

But Win Shares underrates 1B from Konetchy's time. I don't believe it has a bias against shortstops. Shortstops from this era rack up a ton of fielding Win Shares. So I would say that Konetchy should get the edge.

Konetchy was a much better hitter, Sewell was a SS for 8 years and a 3B for 5. I think KJOK is double counting - you can't compare Sewell's offense to other shortstops and then compare Konetchy's offense to firstbasemen - and then on top of that say that Sewell was a good fielding shortstop and Konetchy an excellent firstbaseman and call that a positive for Sewell too.

If you compare their offense based on where they played, then you have to give Konetchy the fielding edge because he was better than his counterparts with the leather (just as Sewell was with the stick).

I have Konetchy slightly ahead, and I am blown away that Sewell finished 4th on this ballot while Bancroft (287 WS, 202 WSaR, .693 PA), was 44th and Konetchy 45th. I just don't get that at all. I think the voters made a mistake, and it wasn't in underrating Konetchy as much as it was in overrating Sewell. The numbers just aren't there.

I was overrating Bancroft earlier and I've since dropped him well off my ballot. I can't see Sewell ever sniffing my ballot. He was a nice player that had three big seasons but it's just not enough. To me, you need a pretty enormous Jennings like peak (or some evidence that there is a systemic bias against you in the system) if you are going to get into the Hall of Merit with just 293 Win Shares.

Actually there may be one thing that I haven't considered, and that could be a career length bonus for middle infielders. Outfielders have longer careers - maybe SS/2B (pre-war 3B) deserve a little boost because their careers were shorter? I don't know. I give something like that for catchers, but I always assumed with infielders their careers were shorter more because by the time they hit their 30s and their D slips, their offense is so bad that you can't put them anywhere else - wheras with catchers it's more of a wear and tear thing.

I just ran the numbers for Joe Tinker - basically the same as Bancroft. 273 WS, 200 WSaR, .685 PA.

Again, how does Sewell just tower over these guys? Do people forget that he played 35% of his career as a 3B?

To add to this point, Sewell played third base after the defensive spectrum shift, when it became an easier position to man, more of the hitter's position we think of it as being now.

It begs the question Why did the Indians switch him? Was it because he simply wasn't good enough to play short, or was it something else? His stats seem to indicate he had lost some speed by age 30. Also, the team did improve a lot in 1929 after the switch.

Here's another - Herman Long. Long had 308 WS when you adjust for season length, 217 WSaR. He had .755 Pennants Added. He had 30 WS seasons in 1891, 1892 and 1893, all in the NL - so you can't say he doesn't have Sewell's peak. Long actually played almost his entire career at shortstop too. His rookie season was in the AA, and he posted 26 WS there, so I could see docking him a touch, but that would just pull him even with Sewell.

How does Long not get a single 15th place vote, while Sewell finishes 4th?

Konetchy was a much better hitter, Sewell was a SS for 8 years and a 3B for 5. I think KJOK is double counting - you can't compare Sewell's offense to other shortstops and then compare Konetchy's offense to firstbasemen - and then on top of that say that Sewell was a good fielding shortstop and Konetchy an excellent firstbaseman and call that a positive for Sewell too.

You are correct (sorf of)!

Sewell has a big edge in position adjusted offense. However, even if they were EVEN in position-adjusted offense, an "average" fielding SS would be more valuable than an "average" fielding 1st baseman because SS still claims more defensive value/responsibility, while on the offensive side every batter is basically 1/9th of the offense.

So I did overstate the difference the way I wrote my reply, but Konetchy's edge in fielding a lesser position still won't get him even with Sewell.

To add to this point, Sewell played third base after the defensive spectrum shift, when it became an easier position to man, more of the hitter's position we think of it as being now.

3rd base was on the move rightward, but it was really not until 1934 that 3rd base "shifted" to the right of 2nd base. Sewell retired in 1933...

It begs the question Why did the Indians switch him? Was it because he simply wasn't good enough to play short, or was it something else? His stats seem to indicate he had lost some speed by age 30. Also, the team did improve a lot in 1929 after the switch.

Can't find exactly why, but the Indians traded George Uhle(!) to the Tigers to get their SS, Jackie Tavener, to play SS for the Indians in 1929. Tavener may have been considered a defensive wiz, but he was certainly a bust with the bat after coming to Cleveland.

Sewell (293 WS, 212 WSaboveReplacement, .733 Pennants Added) Finishes 4th
How does Long (308 WS when you adjust for season length, 217 WSaR. He had .755 Pennants Added) NOT even get a vote?
Joe Tinker - basically the same as Bancroft. 273 WS, 200 WSaR, .685 PA.
Bancroft (287 WS, 202 WSaR, .693 PA) 44th
Konetchy (303 WS, 214 WSaR, .732 PA) 45th

Again, how does Sewell just tower over these guys?
Because he is the shiny new toy-part 2.
Yes, I am angry. I want to know how voters decide to rank the players on their ballot and if they consider the comparable players throughout history. We have had roughly 70-75 years of professional baseball as of this election. There are a heck of a lot more candidates than when this thing started in 1898. So there are comparable players for most every other player.

3rd base was on the move rightward, but it was really not until 1934 that 3rd base "shifted" to the right of 2nd base. Sewell retired in 1933

Yeah, I think I've read something like this, too. Do you know what, if any landmark event prompts this conclusion, that 1934 came along and folks said, "OK, we're there," regarding the 3B defensive shift?

Logically, the shift should have been over by 1929-30, at the peak of the new offensive era. The old school, hit-'em-where-they-ain't, bunt-and-run offensive strategies should have been pretty well buried by then.

Because he is the shiny new toy-part 2.
Yes, I am angry. I want to know how voters decide to rank the players on their ballot and if they consider the comparable players throughout history.

It's a little dismaying to see Sewell pull in 4th, yes. I had thought that we had learned to be conservative with newbies, especially non-obvious choices like Rip, I mean Luke, I mean JOE Sewell.

It is perhaps partly because now we've crossed into the Lively Ball era of "modern baseball." Everything that came before has been superceded, is somehow inferior to Sewell and others. I think enough voters are tending towards this kind of informal timelining that it makes a difference in a close election like 1939.

If Faber is to be used as a precedent, Grimes seems like a no-brainer.

Another note on Sewell. The Sporting News began choosing their major league all-star team after each season beginning in 1925. The picks at SS from 1925 to 1929: Glenn Wright-Joe Sewell-Travis Jackson-Jackson-Jackson. Picks at 3B 1929 to 1933: Pie Traynor-Fred Lindstrom-Traynor-Traynor-Traynor.

So Sewell's fine seasons in 1921 and 1923 had no opportunity to be chosen, but still it doesn't say much for his stature in the eyes of his contemporaries during the bulk of his career.

I will answer some of the questions (although I dont understand the arger out there) about Sewell. I had him 4th on my ballot, and I was being very conservative with him this last year. He will be 1st or 2nd this year. Why is he so high? Simple, he has an 88.4 career Warp 3 and a 45.2 peak Warp3 (peak is best 5 nonconsec). Also his prime, 1921-29, is pretty good too.

Win shares may disagree (and it appears it does) but both career and peak voters who primarily use W3 will have him high on their ballots. Those that combine the 2 (such as myself) will have him especially high.

BTW, what does Lave Cross have to do to buy a vote? Do the career voters really see 15 people out there with a better career?

I understand neither the "arger" nor the anger.

I had Sewell 9th and Bancroft, Tinker, and Long off my ballot. The reason is simple. Sewell was a substantially better hitter.

Here are their career EQA's:

Sewell .280
Bancroft .267
Tinker .261
Long .257

Sewell was a signfificantly better hitter than these guys, a roughly comparable defensive player (although perhaps ranking a very close 3rd or 4th among this list of superb defensive SS's), had a great peak (rivaled only by Long), and played against tougher competition than all of them (though the first half of Long's career may have been tougher). In a field as crowded as the last ballot, the admittedly small difference between Sewell and Bancroft is worth about 10 places and the big hitting gap between Sewell and the other two slots dozens of players between them.

Look at it this way, Dahlen's EQA is .278 and he's a slightly better fielder than Sewell. He played forever and was a no-brainer. Sewell was a roughly comparable player to Dahlen (though I'd take Dahlen if push comes to shove) and played 11 seasons. Doesn't that earn him a mid-ballot slot?

Bobby Wallace played even longer than Dahlen, was also a great SS, and took years to get elected. Why? Because he was only an adequate hitter (.270). Well, Bancroft is to Wallace as Sewell is to Dahlen--very similar player with a much shorter career. If Wallace is somewhere in the lower third of the HoM, then Bancroft is a charter member of the Hall of Very Good.

Because he is the shiny new toy-part 2.

What does this mean, Kelly? Our memories of him when we were kids during the twenties are clouding our judgment of him? Are we transfixed over Sewell's big ears?

We have disagreements over players. Maybe we should learn to respect our reasons for picking certain players over others then accusing others of irrationality.

I will answer some of the questions (although I dont understand the arger out there) about Sewell.

I don't understand it, either. Should I be angry that Carey made it when he wasn't on my ballot? Or how about my still smoldering anger over Joe Kelley's election?

IOW, chill. :-)

On the debate over Sewell:

I'd like to see some data on career length and games per season for starters comparing high-defense infielders to outfielders.

Has anyone put together these numbers?

Andrew, Sewell was a 3B for over a third of his career. He has to be a better hitter than the other guys just to be even with them.

Sewell was a good defensive shortstop. The other 3 were great defensive shortstops. I just don't see the gap.

WARP3
Sewell 88.4
Bancroft 75.4
Tinker 70.8
Long 65.4

WARP1
Tinker 111.6
Bancroft 111.2
Long 107.4
Sewell 103.1

Was the league quality that big of a factor? I don't know. They obviously have no reason to boost the AL like they do, and there are many obvious reasons why the AL would be considered the better league, but I guess I don't see the difference to be so great that it would cause the chasmn between Sewell and the others.

I'd rank them Long/Sewell (could go either way), Bancroft/Tinker, but I'd have it a lot closer.

I agree with Kelly, it's not like these candidates were so weak that Faber and Sewell were obvious choices. I'm buying the Shiny New Toy Theory as well . . .

"What does this mean, Kelly? Our memories of him when we were kids during the twenties are clouding our judgment of him? Are we transfixed over Sewell's big ears?

We have disagreements over players. Maybe we should learn to respect our reasons for picking certain players over others then accusing others of irrationality."

I didn't read Kelly's take as accusing others of being irrational - just that he picked up on a phenomenon where sometimes a new players sneaks by in a weak year without really having to run the guantlet.

I didn't read Kelly's take as accusing others of being irrational - just that he picked up on a phenomenon where sometimes a new players sneaks by in a weak year without really having to run the guantlet.

A "shiny toy effect" suggests some of us are being blinded by other things than statistical facts surrounding a player. If I'm wrong about this, I assume Kelly will correct me. :-)

As for the guantlet, Sewell will have to run it again.

I object to the general suggestion that players, even non-no brainers, should have to "run the gauntlet". Particularly major leaguers; what "new information" is likely to come out in the future on Joe Sewell? We have his entire statistical record, why can't we make a judgment on him now?

I don't agree with it in the case of Negro Leaguers or in the (now past) case of very early baseballers either, but at least that attitude is understandable. Information on them is not generally available to everyone, and there is always more we can learn about these players because the hard evidence is lacking.

But with major leaguers? Should we have downgraded our closest estimates of Sewell (and Faber) because they were new and might actually get elected? Because I sincerely doubt that 5 elections from now, we're going to have any different info on Sewell than we have right now.

I know the consensus is that Sewell was an A- SS while the others were A or A+, but, FYI, here is career FRAA (games played all between 1800 and 1925):

Long 181
Bancroft 106
Sewell 106
Tinker 82

Granted 1/3 of Sewell's games were at 3B, but you can't both count that against him at the bat and in the field.

HOM ELECTEES (min 10 G)
'90-92 - 29 to 30
1893 - 26
'94-03 - 20 to 22
'04-14 - 24 to 27
1915 - 23
1916 - 28
1917 - 22
Then a slide on down from there, at the moment.



P.S.
I think it was last year's ballot discussion which featured a chart which I thought rather clearly demonstrated a tendency for players to be ranked higher in their first year than in subsequent years. And I don't think it can be explained by stronger new candidates, as it just keeps happening.

I suppose one could argue a "fatigue factor" against long-timers instead of shiny new toy, but that's just the flipside of the same phenomenon.
Or did I misread that?

We're showing signs of electing 20s mediocrities over older backlog -- both Carey and Faber were pretty marginal candidates, as were Groh and Covaleski. I'm not saying any of them are unworthy (well, Carey maybe) but you'd think 1 or 2 out of 4 would have been pre-1920 players. If we elect 2 more "moderns" in 1940, I think we can be pretty confident that the process has gone off the rails.

Pike's not on my ballot, but I'd like to see him elected, otherwise supporters of Beckley, Welch, Griffith, Childs, etc. might as well pack up and go home.

Actually there may be one thing that I haven't considered, and that could be a career length bonus for middle infielders. Outfielders have longer careers - maybe SS/2B (pre-war 3B) deserve a little boost because their careers were shorter? I don't know. I give something like that for catchers, but I always assumed with infielders their careers were shorter more because by the time they hit their 30s and their D slips, their offense is so bad that you can't put them anywhere else - wheras with catchers it's more of a wear and tear thing.

I don't see a career bonus for shortstops, at least as compared to first basemen. Let's look at the Top 20 eligible at each position, by total games played (at any position). I apologize if I missed one or two -- I may have, but I don't think it is enough to change the results.

Most Games Played by Players Whose Primary Position Was Shortstop:

1.2792Wagner
2.2670Maranville
3.2443Dahlen
4.2383Wallace
5.2368Davis
6.2200Corcoran
7.2012Peckinpaugh
8.1946Bush
9.1913Bancroft
10.1903Sewell
11.1874Long
12.1825Ward
13.1804Tinker
14.1736Glasscock
15.1728Doolan
16.1710Smith
17.1682Cross
18.1659McBride
19.1659Turner
20.1654Scott


Most Games Played by Players Whose Primary Position Was Shortstop:

1.2523Anson
2.2386Beckley
3.2171Judge
4.2128McInnis
5.2085Konetchy
6.2055Sisler
7.2014Daubert
8.1997Connor
9.1994Tenney
10.1919Chase
11.1872Pipp
12.1866Burns
13.1755Davis
14.1687Tucker
15.1673Brouthers
16.1638Merkle
17.1622Kelly
18.1615Blue
19.1564Doyle
20.1530Fournier

As you can see, pretty much up the line, top shorstops played more games than top first basemen. Is this within a margin or error? Was there more wear and tear on first basemen? In any event, it doesn't warrant giving a career-length bonus to the #10 shortstop as compared to the #5 first baseman.

"Should we have downgraded our closest estimates of Sewell (and Faber) because they were new and might actually get elected? Because I sincerely doubt that 5 elections from now, we're going to have any different info on Sewell than we have right now."

It's not more info Michael, it's more time to digest the info and make sense of it.

It does however justify ELECTING the #2 first baseman.

oops, hit post too early . . .

""Should we have downgraded our closest estimates of Sewell (and Faber) because they were new and might actually get elected? Because I sincerely doubt that 5 elections from now, we're going to have any different info on Sewell than we have right now."

It's not more info Michael, it's more time to digest the info and make sense of it. "

Not just more time to make sense of it, but time to have your opinion change one way or the other. I don't necessarily support the same candidates I did 6 months ago. It took me awhile to see how good Pike was for example. I recently did some revamping and dropped Hughie Jennings down a bunch, etc.. When a non-no-brainer rides in on his first try, he doesn't get that chance to bounce around, have his numbers poked and prodded, etc..

I'm not saying we should consciously downgrade all new candidates, but there's something to be said for being conservative and taking a little time to let the numbers sink in.

Good stuff PhillyBooster - I was just trying to find something that could have me rate Sewell higher than Konetchy. When this many people disagree with me, I tend to try to figure out why I must be wrong.

I guess with Sewell it boils down to league quality as the only thing that separates him from the others.

As an alternative to the FRAA list, here are the defensive Win Shares numbers for the players in question. They aren't adjusted seasonally, that would take too much time right now. I took career fielding WS, divided by total WS and multiplied by total season length adjusted win shares:

Herman Long 121.7
Joe Tinker 120.1
Dave Bancroft 110.9
Joe Sewell 94.8

As you can see, pretty much up the line, top shorstops played more games than top first basemen. Is this within a margin or error? Was there more wear and tear on first basemen? In any event, it doesn't warrant giving a career-length bonus to the #10 shortstop as compared to the #5 first baseman.

Honus Wagner ended up at 1B, so did Hughie Jennings. This is more important to me than actual career length. Were players at other positions shifted to 1B in this era to lengthen their careers (either because they could still hit, or for loyalty reasons).

In today's game, many first basemen don't age well because the athletic limitations that keep them from playing other positions magnify as they get older... not necessarily due to wear and tear. (See Kent Hrbek & Boog Powell). I'm not saying this necessarily applies to the 1900s and 1910s.

Phillybooster,

Re post 81:

Do you have similar data for outfielders?

I could certainly see, in a period of small gloves, 1B having more wear and tear than SS. Not that this was the case necessarily, but it's easy to construct an argument . . . there was lots of hitting and running, and combining that with few baserunners meant fewer DP pivots to get clobbered on, right? And 1B were still taking a ton of throws, which had to be rough on the hands. Just a theory . . .

Here are the left fielders. They look similar to the first basemen, and less prolific than the shortstops, as well. These are total games, not just games in LF.

1Wheat2410
2Clarke2242
3Sheckard2122
4Magee2087
5Burkett2066
6O'Rourke1999
8Burns1853
7Kelley1853
9Delahanty1835
10Veach1821
11Jamieson1779
12Anderson1635
13Selbach1610
14Mann1498
15Stovey1486
16Lewis1459
17Walker1421
18BMeusel1407
19Graney1402
20Williams1397

Honus Wagner ended up at 1B, so did Hughie Jennings. This is more important to me than actual career length. Were players at other positions shifted to 1B in this era to lengthen their careers (either because they could still hit, or for loyalty reasons).

In fact, if you look at just games at SS, the shorstops look a lot like the 1B and LF players for ALL games. Sewell -- with 1216 games -- is 21st on this list.

I don't really have a good explanation, but the fact is that top shortstops simply did NOT have shorter careers. If anything, they were longer.

2153Maranville
2132Dahlen
2073Corcoran
1982Peckinpaugh
1887Wagner
1873Bancroft
1826Wallace
1794Long
1743Tinker
1734Bush
1676Cross
1665Smith
1643Scott
1628Glasscock
1626McBride
1625Doolan
1564McKean
1448Fletcher
1447Gerber
1372Davis

Honus Wagner ended up at 1B, so did Hughie Jennings. This is more important to me than actual career length. Were players at other positions shifted to 1B in this era to lengthen their careers (either because they could still hit, or for loyalty reasons).

I agree. I also think there is a problem creating position lists when all eras are not equal. Comparing a player to his contemporaries makes more sense, IMO.

But if we're going to use position lists, maybe we should set them up using career WS (plus career WS per 162 Games) or WARP. Then we can see that shortstops fall short of many of the other positions because of the postion's wear-and-tear on the body.

We're showing signs of electing 20s mediocrities over older backlog -- both Carey and Faber were pretty marginal candidates

Faber in a lot of ways is the same player as Welch. Long career for his time with a fine but not truly outstanding peak. In fact, I had Welch and Faber on the ballot contiguously. :-)

Max Carey had a fairly long career, but there are CF with longer careers. He was clearly a brilliant defender for a very long time. He was a somewhat useful offensive player. His career rate stats are dragged down becuase his defense kept him in the lineup when his bat wouldn't have. But even at his best, we had a number of not-yet-elected CF who were clearly better hitters than Carey. We elected him.

Rabbit Maranville had one of the longest careers ever for a SS. He was clearly a brilliant defender for a very long time. True, he wasn't much on offense, but there's a little of the same effect as for Carey; his glove keeping him in the lineup after his offense had slipped. Maranville appeared on only 7 ballots, none higher than 9th.

I'm not saying the two cases are exactly equivalent, and I personally had Carey higher than Maranville, which put Carey on my ballot and Maranville off. But there are some parallels.

Shortstops from this era rack up a ton of fielding Win Shares.

This is not true. The way Win Shares is constructed makes it close to impossible for this to be true. On average, shortstops get 18% of the Fielding win shares. On average, fielders get 32.5% of the Defensive Win Shares. On average, defense gets 52% of the total Win Shares. After adjusting to 162 games, an average SS in any era gets close to the same number of fielding Win Shares, no matter how many ground-balls vs fly balls are hit, with a small (some would say inadequate) correction for strikeouts.

The GOOD shortstops from this era may rack up a ton of fielding Win Shares compared to today, but that is because there are also BAD shortstops in this era that get less. The variance is higher; IOW, the fielding replacement level is markedly lower.

The case for Sewell for me comes down to his comparison vs. his contemporaries:

1. Dave Bancroft
2. Travis Jackson
3. Not much else....Glenn Wright, Charlie Hollocher, Topper Rigney, Wally Gerber, Mark Koening?

Sewell gave his team TREMENDOUS value as he was almost always, by far, the best SS on the field, offensively for sure, and usually defensively too.

I disagree Jim.

WS has some things built in to give more fielding WS in eras with low strikeout totals. This factor isn't as strong as it should be, but it is there. Strikeout totals were very low in this era, so fielding WS are higher all-around. Of course it shows up the most in the positions that get the most WS, like SS.

KJOK - if ARod were transported back to the 70s, he'd be better relative to his peers at SS, but that's only becuase it was a weak era for shortstops. I don't see how that makes him any more valuable as a player. Sewell played SS from 1920-28. Expand his era by 15 years on either end and now he's being compared to Honus Wagner and Arky Vaughn and Luke Appling. By only comparing him to SS that played in the exact 9 year period he played, you are loading the sample by using selective endpoints.

Joe - It's not a selective endpoint argument in that THOSE ARE THE YEARS HE ACTUALLY PLAYED. Wagner and Vaughan and Appling weren't in the league. They should be compared to THEIR PEERS to see how they stack up.

If Arod played in the 70's, he might have hit .270 with 10 HR's. Or more likely, he would have been moved to CF....

I disagree that they should strictly be compared to their peers. Why should a player get a bonus if his peers were going through a cyclical downcycle for the few years he played?

We are comparing players across history to each other.

It's one thing if there is something unique to an era (like playing 1B without a glove or something) that makes it a different position. But giving a player a bonus because his position was going through a down cycle, or a demerit for playing in an era that happened to be strong at the position makes no sense to me at all.

How do you know it was a "down" cycle? It may just appear to be a down cycle due to the nature of the era.

It doesn't make it hard to compare players across eras. If Joe Sewell was 50% better than his peers, and Derek Jeter(?) was 48% better, then Sewell was more valuable, and he should rank higher.